The numpy model of computation in Python has proven to be one of the most successful ways to integrate high-performance computational code into an application. This talk offers a foundational conceptualization for this approach and discusses its strengths and limitations.